TinkerThe split-order quote, about to be counted.
Thinking recorded
Run a question to see what the model wrote before it answered.
Chapter 2 of 7
Thinking is tokens
Thinking fixed the split-order quote. The support lead’s next question is the one every team asks: what does it cost, and how long does the customer wait?
This chapter: Where does thinking show up on the bill and on the clock?
You will learn
- Why thinking is counted in tokens, like the answer
- Why thinking is billed as output, even when you never see it
- Why every thinking token is written before the first word of the answer
Words you’ll meet (3)
- Token
- A small piece of text a model reads or writes; prices and limits are counted in them. The Tokens module covers them.
- Output tokens
- The tokens a model writes, billed separately from the tokens it reads. Thinking counts as output.
- Wait
- How long the customer waits for the whole answer: the thinking is written first, then the answer.
The picture on the left is live. Each step below asks you to do one thing in it; the story continues here as you go.
Step 1 of 4
Thinking is text, so it is made of tokens: the small pieces a model reads and writes, which prices and limits are counted in. Before you count them: